CN112365935B - Cement free calcium soft measurement method based on multi-scale depth network - Google Patents
Cement free calcium soft measurement method based on multi-scale depth network Download PDFInfo
- Publication number
- CN112365935B CN112365935B CN202011125721.7A CN202011125721A CN112365935B CN 112365935 B CN112365935 B CN 112365935B CN 202011125721 A CN202011125721 A CN 202011125721A CN 112365935 B CN112365935 B CN 112365935B
- Authority
- CN
- China
- Prior art keywords
- soft measurement
- cement
- variable
- auxiliary
- free calcium
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/60—In silico combinatorial chemistry
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C60/00—Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Abstract
The invention discloses a cement free calcium soft measurement method based on a multi-scale depth network, belonging to the technical field of cement clinker quality soft measurement detection, and the concrete method comprises the following steps: analyzing a cement production process, selecting process variables related to the f-CaO content in cement clinker, and determining auxiliary variables required by a soft measurement model; adopting a Lauda criterion to mark an abnormal value in each auxiliary variable, and replacing the abnormal value and the missing value in each auxiliary variable by using a mean value of the auxiliary variable; carrying out 3-layer wavelet packet decomposition on the auxiliary variable and extracting real-time characteristics; sending the extracted real-time characteristics into an LSTM model, training the model, and correcting model parameters through an error back propagation algorithm; and (5) predicting the f-CaO content by using a trained LSTM model. The method can extract more variable characteristics, can more accurately predict the free calcium value in the cement clinker, and has practical guiding significance for cement production.
Description
Technical Field
The invention relates to the technical field of cement clinker quality soft measurement detection, in particular to a cement free calcium soft measurement method based on a multi-scale depth network.
Background
The content of free calcium oxide (f-CaO) in the cement clinker is an important index for measuring the quality condition of the cement. The content of f-CaO is closely related to the stability of cement products, the strength of cement clinker and the energy consumption of production. Too high f-CaO content leads to a decrease in the quality of the cement clinker, while too low f-CaO content leads to an increase in energy consumption. In the current cement production enterprises, the content value of f-CaO in cement clinker is mainly determined by adopting a method of sampling on site and then carrying out manual assay. However, the f-CaO content result obtained by off-line test has serious hysteresis for guiding cement production, and real-time control of the cement production process is difficult to realize. Meanwhile, the process variables also have the characteristics of large inertia, multi-coupling and the like in the cement production process, so that an accurate f-CaO content prediction model is difficult to establish. With the development of soft measurement technology, data-driven soft measurement modeling technology is applied to f-CaO content prediction. Zhao Peng Cheng et al proposed a soft measurement method of f-CaO content of cement clinker by combining the least square method with a support vector machine. But since only 5 process variables are selected as auxiliary variables, overfitting is easily caused when the sample size is too large. WanC et al established a soft measurement model of f-CaO content of cement clinker by using a sliding window recursive least square method. Although the method improves the real-time property of f-CaO content measurement, the prediction precision is greatly reduced when the production working condition is changed. The cement clinker production is a complex and variable process, and if a high-precision soft measurement model with strong interference resistance is to be obtained, not only process variables influencing the clinker quality in the production flow need to be fully considered, but also important characteristics influencing the f-CaO content in process variable data need to be fully extracted.
Disclosure of Invention
Aiming at the existing problems, the invention provides a cement free calcium soft measurement method based on a multi-scale depth network.
In order to achieve the purpose, the invention is realized by the following technical scheme:
a cement free calcium soft measurement method based on a multi-scale depth network comprises the following steps:
step 1: auxiliary variables related to the f-CaO content of the cement clinker are selected.
By analyzing the production flow of the cement clinker and the mechanism of f-CaO generation, 10 process variables related to the f-CaO content are selected as auxiliary variables of the soft measurement model.
Step 2: and (4) preprocessing data of auxiliary variables.
The outliers and missing values in the 10 auxiliary variables in step 1 are first labeled using the ralda criterion and the mean of the auxiliary variables is used for the outliers and missing values in each auxiliary variable. In order to avoid the influence of the non-uniformity of the auxiliary variable dimension on the prediction result, the data of all 10 auxiliary variables are normalized.
And step 3: and carrying out wavelet packet decomposition on the auxiliary variable and extracting real-time characteristics.
And segmenting each auxiliary variable, then carrying out wavelet packet decomposition on each segment in each variable, calculating the energy of each node after decomposition, and taking the node with larger energy as the characteristic of the variable in the time period.
And 4, step 4: and (4) sending the real-time features extracted in the step (3) into an LSTM model and training the model.
Each sample used in the LSTM model training consists of the features extracted in step 3 and the f-CaO content values. The network is forward trained using randomly initialized parameters and adjusted using a back propagation algorithm.
And 5: and (5) predicting the f-CaO content by using a trained LSTM model.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention adopts a soft measurement mode to establish the cement clinker free calcium soft measurement method, and can effectively reduce the free calcium measurement cost.
2. The invention uses the characteristics of the auxiliary variable in a period of time as the input of the soft measurement model, and can effectively improve the anti-interference capability of the model.
3. The cement clinker free calcium soft measurement model established by the invention can detect the content of free calcium in real time and can provide certain guidance for an operator to control the cement production process.
Drawings
FIG. 1 is a diagram of a soft measurement scheme for f-CaO content of cement clinker provided by the present invention;
FIG. 2 is an exploded view of a three-layer wavelet packet signal used in the present invention;
FIG. 3 is a diagram of a recurrent neural network in accordance with an embodiment of the present invention;
FIG. 4 is a block diagram of a long term short term memory neural network according to an embodiment of the present invention;
FIG. 5 is a diagram of the predicted result of the method of the present invention.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
the invention provides a modeling method of a soft measurement model of f-CaO content of cement clinker by combining wavelet packet decomposition and an LSTM neural network, and a designed modeling scheme is shown as an attached drawing 1. Firstly, analysis is carried out according to the production process, and a process variable related to the f-CaO content is selected as an auxiliary variable used for soft measurement modeling. Outliers are labeled according to the Lauda rule, and the mean of the variable is used instead of outliers and missing values in the variable. And segmenting each auxiliary variable, and performing 3-layer wavelet packet decomposition on each segment. The wavelet packet 3 layer decomposition is shown in figure 2. And calculating the node energy after each variable is decomposed according to the energy function, and taking the node with higher energy as the characteristic of the variable in the time period. And inputting a sample consisting of the characteristics and the f-CaO content value into an LSTM neural network for training. And finally, using the trained model for f-CaO content prediction. The method comprises the following steps:
step 1: and analyzing the cement production process and determining auxiliary variables required by the soft measurement model.
According to the cement production process, cement clinker is a solid particulate matter obtained by burning raw materials mixed with limestone, clay and the like at a high temperature to melt, cooling and crushing. Among them, calcium oxide which is not combined in the process of firing cement clinker is called free calcium oxide (f-CaO). The stability of the cement finished product is influenced by the content of the f-CaO. The f-CaO content is therefore generally a key indicator of the quality of the cement clinker. In the cement clinker production process, raw materials are firstly preheated in a preheater, the preheated raw materials are sent into a pyrolysis furnace for heat absorption decomposition of carbonate, and the process mainly comprises the coal feeding amount of the pyrolysis furnace, the coal feeding amount of a kiln head and secondary air recycled into a rotary kiln from a grate cooler for providing high temperature. The material then enters the rotary kiln and the carbonate that is not decomposed in the decomposing furnace will be further decomposed in the rotary kiln. The high-temperature fan and the EP fan enable huge air pressure difference to be generated in the kiln, and guarantee the smoothness of the air path of the cement burning system. In order to ensure that the materials in the rotary kiln can be uniformly heated to react, the rotary kiln is driven by a motor to rotate, and the current of a kiln main machine can change along with the viscosity of the materials in the kiln. Finally, the high-temperature clinker coming out of the rotary kiln is rapidly cooled by the grate cooler and then flows out from the feed opening
From the above analysis, 10 variables closely related to the f-CaO content of the clinker were selected: feeding amount 1 feedback, decomposing furnace coal feeding amount feedback, high-temperature fan rotating speed feedback, kiln tail temperature, decomposing furnace outlet temperature, kiln current feedback, pressure feedback under a two-chamber grate, secondary air temperature feedback, kiln head negative pressure feedback and kiln head coal feedback.
Step 2: the outliers in each auxiliary variable are marked using the Lauda criterion, and the outliers and missing values in each auxiliary variable are replaced with the mean of the auxiliary variable.
The Lauda criterion is shown in formula (1):
|x i -μ|≥3σ (1)
in the formula (1), x represents an auxiliary variable, x i Represents the ith value of the auxiliary variable, μ represents the mean of the auxiliary variable, and σ represents the standard deviation of the auxiliary variable.
And step 3: and 3-layer wavelet packet decomposition is carried out on the auxiliary variable and real-time features are extracted.
As the acquisition frequency of the auxiliary variables is faster relative to the acquisition frequency of the f-CaO content, the data of each auxiliary variable are segmented according to the acquisition frequency of the f-CaO content, and finally a segment of time series data of each auxiliary variable corresponding to one f-CaO content value is formed. And (3) taking a time sequence of each auxiliary variable as an original signal to carry out 3-layer wavelet packet decomposition and explaining a characteristic extraction process of the wavelet packet decomposition by combining with the figure 2. Aiming at an original signal S (t), a set of low-pass and high-pass conjugate orthogonal filter coefficients { h } k } k∈Z And
the coefficients of the wavelet packet transform of each layer can be recurred by equation (2).
In the formula (2), n is a frequency index, k is a position index, and j represents the number of layers subjected to wavelet packet decomposition. P s And (n, j, k) represents a coefficient sequence obtained by wavelet packet transformation of the original signal. As can be seen from fig. 2, the original signal is decomposed by 3 layers of wavelet packets to obtain 8 coefficient sequences, which are respectively represented by SSS, dSS, SdS, ddS, SSd, dSd, Sdd, and ddd. Then, the energy intensities of the 8 coefficient sequences are respectively calculated through the constructed energy functions, and the first 4 coefficient sequences with higher energy intensities are extracted as real-time characteristics of the original signal in a period of time. The constructed energy function is shown as formula (3):
wherein x i Representing the value of the ith element in the coefficient sequence. And performing 3-layer wavelet packet decomposition and feature extraction on the time sequence segment of each auxiliary variable, and taking the feature of each variable in a time segment as the input of the soft measurement model.
And 4, step 4: and (4) establishing an LSTM neural network model, and training by using the characteristics obtained in the step (3).
The LSTM neural network is fully called a long-term and short-term memory neural network. Is an improved form of recurrent neural networks. In the drawings, FIG. 3 is a structural diagram of a Recurrent Neural Network (RNN), and FIG. 4 is a structural form of an LSTM neural network. The following discussion describes the use and deployment of the LSTM neural network with respect to fig. 3 and 4.
RNNs are a form of neural networks. RNNs can persist information by connecting hidden layer neurons. This property makes RNNs suitable for modeling time series. In the context of figure 3, it is shown,arrows indicate connections between neurons, A hidden layer nerve units, X t Representing inputs of a neural network, h t Representing the output of the hidden layer unit. In standard RNN structures, a generally uses a tanh function of the form shown below:
RNN can predict current information using past information, but the effect of RNN is not ideal when the time interval between related information and information to be predicted becomes very large. The LSTM neural network model is thus proposed, and LSTM effectively remedies this disadvantage of RNN by introducing a gate structure. LSTM does not change the chain structure of RNN, but changes the hidden unit form of RNN. The LSTM adds a gate control unit and a cell state (cell state) to the hidden layer unit, so that the whole network has a function of selectively saving or forgetting information. The established LSTM network prediction model is explained by fig. 4 and fig. 5. In LSTM, the cellular state corresponds to a pathway for information transfer, allowing information to be transferred between layer neurons. And the addition and removal of information are simultaneously controlled by three door control units, namely a forgetting door, an input door and an output door. Wherein:
forgetting gates are used to determine which useful data to delete or retain. The unit inputs the output information of the previous hidden layer and the current input information into a sigmoid function at the same time to achieve the purpose of control. The sigmoid function expression is shown in formula 4:
the control expression of the forgetting gate is shown in formula 6:
f t ＝σ(W f ·[h t-1 ,x t ]+b f ) (6)
wherein W f Is the weight matrix of the forgetting gate, b f Is a biased term for a forgetting gate.
The input gate functions to update the cell state. The current input information and the output information of the last hidden layer unit are simultaneously input into a sigmoid function and a tanh function, and then the output values of the two functions are multiplied. The operation result is used to update the cell state. The control expression of the input gate is shown in equations 7 and 8:
i t ＝σ(W i ·[h t-1 ,x t ]+b i ) (7)
in the above formula W i Is the weight matrix of the input gate, W c Is a weight matrix of the state of the computational cell, b i Is a bias term that is input to the gate,
representing the currently entered cell state.
The output gate functions to determine the value of the next hidden state. Firstly, simultaneously inputting the output information of the previous hidden layer unit and the current input information into a sigmoid function, then inputting the new cell state into a tanh function, finally, performing multiplication operation on the output values of the two functions, taking the final result as the current hidden layer state, and inputting the final result and the current cell state into the next hidden layer unit. The control expression of the input gate is shown in equations 9 and 10:
o t ＝σ(W o ·[h t-1 ,x t ]+b o ) (9)
h t ＝o t *tanh(C t ) (10)
in the above formula, W o Is an output matrix, b o Is the output offset term, h t Is the final output value of the LSTM cell, and
the training mode of the LSTM network adopts error back propagation training. The mean square error function (MSE) is chosen as the loss function E, and the MES equation is shown in equation 11:
wherein y is i Representing the true value of clinker free calcium, y i ' denotes the model predicted clinker free calcium output, n denotes the total number of training samples.
The error back propagation of LSTM includes two directions: one is backward propagation in time and one is to propagate the error up one layer. The parameters to be trained for LSTM are: w f 、b f 、W i 、b i 、W c 、b c 、W o 、b o . Wherein the actual weight matrices to be updated are 8 in number, since each weight matrix connects two different neural units. Respectively as follows: w is a group of fx 、W fh 、W ix 、W ih 、W cx 、W ch 、W ox 、W oh 。
The propagation of the error to the previous time instant is calculated as follows:
at time t, the LSTM output value is h t . Then the error term δ at time t is defined t As shown in formula 12 below:
four weighted inputs f that simultaneously define the LSTM t 、i t 、
o t And their corresponding error terms are respectively shown as follows:
net f,t ＝W fh h t-1 +W fx x t +b f (13)
net i,t ＝W ih h t-1 +W ix x t +b i (14)
net o,t ＝W oh h t-1 +W ox x t +b o (16)
the back propagation process of the error in time is to calculate the value delta of the error term at the moment t-1 t-1 . The calculation formula is as follows:
the error propagation to the upper layer is calculated as follows:
if the current layer is L layer, the error term delta of the previous layer (L-1 layer) t L-1 Is the inverse of the weighted input of the loss function to the L-1 layer, and is calculated as follows:
the parameter update calculation formula in the LSTM is as follows:
W fh the gradient updates are as follows:
wherein
W ih The gradient updates are as follows:
wherein
W ch The gradient update is as follows:
wherein
W oh The gradient update is as follows:
wherein
b f The bias term is updated as follows:
b i the bias term is updated as followsFormula (II):
b c the bias term is updated as follows:
b o the bias term is updated as follows:
W fx the gradient update is as follows:
W ix the gradient update is as follows:
W cx the gradient updates are as follows:
W ox the gradient update is as follows:
and D, storing the finally trained model for prediction in the step five.
And 5: and (5) predicting the f-CaO content by using a trained LSTM model.
The prediction results of the model are shown in fig. 5.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solution of the present invention by those skilled in the art should fall within the protection scope defined by the claims of the present invention without departing from the spirit of the present invention.
Claims (3)
1. A cement free calcium soft measurement method based on a multi-scale depth network is characterized by comprising the following steps: the method comprises the following steps:
step 1: analyzing a cement production process, and selecting process variables related to the f-CaO content in cement clinker so as to determine auxiliary variables required by a soft measurement model;
and 2, step: adopting a Lauda criterion to mark an abnormal value in each auxiliary variable, and replacing the abnormal value and the missing value in each auxiliary variable by using a mean value of the auxiliary variable;
and step 3: carrying out 3-layer wavelet packet decomposition on the auxiliary variables and extracting real-time characteristics, segmenting each auxiliary variable, then carrying out wavelet packet decomposition on each segment in each variable, calculating the energy of each node after decomposition, and taking a plurality of nodes with larger energy as the characteristics of the variable in the time period;
the acquisition frequency of the auxiliary variables is faster relative to the acquisition frequency of the f-CaO content, so that the data of each auxiliary variable is segmented according to the acquisition frequency of the f-CaO content, and finally a section of time sequence data of each auxiliary variable corresponding to one f-CaO content value is formed; decomposing the auxiliary variable by using 3 layers of wavelet packet decomposition, and extracting the features by using the constructed energy function; aiming at an original signal S (t), a set of low-pass and high-pass conjugate orthogonal filter coefficients { h } k } k∈z And { g } k } k∈z Then, the coefficients of the wavelet packet transform of each layer can be obtained by recursion of equation (2):
in the formula (2), n is a frequency index, k is a position index, j represents the number of layers subjected to wavelet packet decomposition, and P s (n, j, k) represents a coefficient sequence obtained by wavelet packet transformation of an original signal; after 3 layers of wavelet packet decomposition, 8 coefficient sequences are obtained in total from the original signal, SSS, dSS, Sds, ddS, SSd, dSd, Sdd and ddd are used for representing the original signal respectively, then the energy intensity of the 8 coefficient sequences is calculated respectively through the constructed energy function, and the first 4 coefficient sequences with higher energy intensity are extracted as the real-time characteristics of the original signal in a period of time; the constructed energy function is shown in formula (3):
wherein x i Representing the value of the ith element in the coefficient sequence; 3-layer wavelet packet decomposition is carried out on the time sequence segment of each auxiliary variable, characteristics are extracted, and the characteristics of each variable in a time segment are used as the input of a soft measurement model;
and 4, step 4: sending the real-time characteristics extracted in the step 3 into an LSTM model, training the model, and correcting the model parameters through an error back propagation algorithm;
and 5: and (5) predicting the f-CaO content by using a trained LSTM model.
2. The cement free calcium soft measurement method based on the multi-scale depth network is characterized in that: in step 1, 10 process variables closely related to free calcium oxide are selected as auxiliary variables of a soft measurement model of the free calcium by analyzing the mechanism of the generation of the free calcium oxide in clinker and related influencing factors.
3. The cement free calcium soft measurement method based on the multi-scale depth network as claimed in claim 2, characterized in that: in step 1, the 10 modeling auxiliary variables selected are: feeding amount 1 feedback, decomposing furnace coal feeding amount feedback, high-temperature fan rotating speed feedback, kiln tail temperature, decomposing furnace outlet temperature, kiln current feedback, pressure feedback under a two-chamber grate, secondary air temperature feedback, kiln head negative pressure feedback and kiln head coal feedback.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011125721.7A CN112365935B (en) | 2020-10-20 | 2020-10-20 | Cement free calcium soft measurement method based on multi-scale depth network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011125721.7A CN112365935B (en) | 2020-10-20 | 2020-10-20 | Cement free calcium soft measurement method based on multi-scale depth network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112365935A CN112365935A (en) | 2021-02-12 |
| CN112365935B true CN112365935B (en) | 2022-08-30 |
Family
ID=74507399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011125721.7A Active CN112365935B (en) | 2020-10-20 | 2020-10-20 | Cement free calcium soft measurement method based on multi-scale depth network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112365935B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113593657B (en) * | 2021-07-26 | 2024-01-09 | 燕山大学 | Cement free calcium soft measurement system with quality target as guiding semi-supervised learning |
| CN114994294B (en) * | 2022-05-20 | 2023-04-18 | 燕山大学 | Soft measurement method for free calcium of cement clinker based on attention and window gating mechanism |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104723171A (en) * | 2015-03-17 | 2015-06-24 | 洛阳理工学院 | Cutter wear monitoring method based on current and acoustic emission compound signals |
| CN107122802A (en) * | 2017-05-02 | 2017-09-01 | 哈尔滨理工大学 | A kind of fault detection method based on the rolling bearing for improving wavelet neural network |
| CN110000610A (en) * | 2019-04-17 | 2019-07-12 | 哈尔滨理工大学 | A kind of Tool Wear Monitoring method based on multi-sensor information fusion and depth confidence network |
| CN110456217A (en) * | 2019-08-08 | 2019-11-15 | 辽宁工程技术大学 | A kind of MMC Fault Locating Method based on WPD-FOA-LSSVM dual model |
| CN111104891A (en) * | 2019-12-13 | 2020-05-05 | 天津大学 | Composite characteristic optical fiber sensing disturbing signal mode identification method based on BiLSTM |
| CN111680661A (en) * | 2020-06-19 | 2020-09-18 | 哈尔滨工业大学 | Mechanical rotating part performance degradation tracking method based on multi-feature fusion |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105559777B (en) * | 2016-03-17 | 2018-10-12 | 北京工业大学 | Electroencephalogramrecognition recognition method based on wavelet packet and LSTM type RNN neural networks |
| CN109147878B (en) * | 2018-10-08 | 2021-10-15 | 燕山大学 | Soft measurement method for free calcium of cement clinker |
| CN110132565A (en) * | 2019-05-28 | 2019-08-16 | 河海大学 | A kind of OLTC method for diagnosing faults combined based on wavelet packet and neural network |
| CN110610035B (en) * | 2019-08-28 | 2022-12-16 | 武汉科技大学 | Rolling bearing residual life prediction method based on GRU neural network |
| CN111652348B (en) * | 2020-02-29 | 2023-09-01 | 上海电力大学 | Power battery pack fault fusion diagnosis method and system based on improved CNN |
-
2020
- 2020-10-20 CN CN202011125721.7A patent/CN112365935B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104723171A (en) * | 2015-03-17 | 2015-06-24 | 洛阳理工学院 | Cutter wear monitoring method based on current and acoustic emission compound signals |
| CN107122802A (en) * | 2017-05-02 | 2017-09-01 | 哈尔滨理工大学 | A kind of fault detection method based on the rolling bearing for improving wavelet neural network |
| CN110000610A (en) * | 2019-04-17 | 2019-07-12 | 哈尔滨理工大学 | A kind of Tool Wear Monitoring method based on multi-sensor information fusion and depth confidence network |
| CN110456217A (en) * | 2019-08-08 | 2019-11-15 | 辽宁工程技术大学 | A kind of MMC Fault Locating Method based on WPD-FOA-LSSVM dual model |
| CN111104891A (en) * | 2019-12-13 | 2020-05-05 | 天津大学 | Composite characteristic optical fiber sensing disturbing signal mode identification method based on BiLSTM |
| CN111680661A (en) * | 2020-06-19 | 2020-09-18 | 哈尔滨工业大学 | Mechanical rotating part performance degradation tracking method based on multi-feature fusion |
Non-Patent Citations (1)
| Title |
|---|
| Fault Diagnosis and Tolerance Control of Five-Level Nested NPP Converter Using Wavelet Packet and LSTM;Shu Ye et al;《IEEE TRANSACTIONS ON POWER ELECTRONICS》;20200228;第1-15页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112365935A (en) | 2021-02-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112365935B (en) | Cement free calcium soft measurement method based on multi-scale depth network | |
| CN110378044B (en) | Multi-time scale convolution neural network soft measurement method based on attention mechanism | |
| CN109147878B (en) | Soft measurement method for free calcium of cement clinker | |
| CN111950191B (en) | Rotary kiln sintering temperature prediction method based on hybrid deep neural network | |
| CN106202946A (en) | Clinker free calcium levels Forecasting Methodology based on degree of depth belief network model | |
| CN113190974B (en) | Roller kiln temperature field multipoint prediction method based on deep learning | |
| CN113591374B (en) | Cement finished product specific surface area prediction method based on convolution simple circulation unit network | |
| CN110070217A (en) | A kind of Forcasting Sinter Quality method of Kernel-based methods parameter | |
| CN109685283B (en) | Method for predicting submerged arc furnace working condition based on confidence rule base reasoning | |
| Li et al. | Dynamic time features expanding and extracting method for prediction model of sintering process quality index | |
| CN109342703B (en) | Method and system for measuring content of free calcium in cement clinker | |
| CN107038307A (en) | Mechanism predicts integrated modelling approach with the Roller Conveying Kiln for Temperature that data are combined | |
| CN113589693B (en) | Cement industrial decomposing furnace temperature model predictive control method based on neighborhood optimization | |
| CN111950697A (en) | Cement product specific surface area prediction method based on gated cycle unit network | |
| Wu et al. | Neural-network-based integrated model for predicting burn-through point in lead–zinc sintering process | |
| CN105574297A (en) | Self-adaption blast-furnace melt silicon content tendency forecasting method | |
| CN112906976A (en) | Cement clinker free calcium sample data enhancement and prediction method based on R-WGAN | |
| Hao et al. | R-WGAN-based multitimescale enhancement method for predicting f-CaO cement clinker | |
| CN111128312B (en) | Zinc oxide volatilization kiln mixed modeling method based on mechanism and support vector machine | |
| Wang et al. | RBF neural network soft-sensor modeling of rotary kiln pellet quality indices optimized by biogeography-based optimization algorithm | |
| Zhang et al. | Prediction of compressive strength of geopolymer concrete based on support vector machine and modified cuckoo algorithm | |
| CN111984919B (en) | Method and system for calculating temperature field and material density of rotary cement kiln | |
| CN115392125A (en) | Temperature prediction method for rotary cement kiln | |
| Liu et al. | Thermal efficiency prediction model of cement clinker production based on fuzzy c-means monitoring clustering | |
| Zhou et al. | Terminal Temperature Prediction of Molten Steel in VD Furnace based on XGBoost and LightGBM Algorithms |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |